Aller au contenu
ModelPrésentéExterne
Browse files

Qwen3 8B INT8

INT8 (W8A8) quantized version of Qwen/Qwen3-8B, created using llm-compressor with calibrated quantization.

Sourcehuggingface.co/vistralis/Qwen3-8B-INT8/tree/26c4df0d86f4a385b63495cf8b805535e8e62133
TypeOtherModèle de baseQwen/Qwen3-8BVersionRevision 26c4df0BClasse

Source du modèle

Type
Other
Taille du fichier
4,18–4,61 GB
Modèle de base
Qwen/Qwen3-8B
Version
vRevision 26c4df0
Mis à jour
9 oct. 2026
Téléchargements
52
Source
Hugging Face

Description de la source

Qwen3-8B-INT8

INT8 (W8A8) quantized version of Qwen/Qwen3-8B, created using llm-compressor with calibrated quantization.

Overview

PropertyValue
Base ModelQwen/Qwen3-8B
Parameters8.19B
QuantizationINT8 (W8A8)
Formatcompressed-tensors
Toolllm-compressor
Disk Size~9.4 GB (2 shards)

Intended Use

Quantized text encoder for Flux 2 Klein 9B image generation pipelines. Architecturally identical to the Klein 9B text encoder.

Quantization Details

  • Scheme: W8A8 — 8-bit integer weights and activations
  • Targets: All Linear layers (excluding lm_head)
  • Calibration: 256 samples from C4, sequential pipeline with CPU offloading

Hardware Requirements

  • Minimum: Any CUDA GPU with INT8 tensor core support
  • Fallback: Dequantizes to BF16 on unsupported hardware

Sources

1 source
Hugging FacePrincipalVersion 26c4df0d86f4a385b63495cf8b805535e8e62133
huggingface.co/vistralis/Qwen3-8B-INT8/tree/26c4df0d86f4a385b63495cf8b805535e8e62133

Vérifié 9 oct.

Download files

Artefacts du modèle

2 artefacts

model-00001-of-00002.safetensors

safetensors · 4,61 GB · SHA-256 23f369c3b81e…1fcd · Hugging Face

Download shard

model-00002-of-00002.safetensors

safetensors · 4,18 GB · SHA-256 26c0d77f150c…37c4 · Hugging Face

Download shard

Extraits de sources

3 extraits
Readme31 juil.huggingface.co/vistralis/Qwen3-8B-INT8

--- base_model: Qwen/Qwen3-8B library_name: transformers tags: - quantized - int8 - compressed-tensors - llm-compressor - flux2 - text-encoder license: apache-2.0 --- # Qwen3-8B-INT8 INT8 (W8A8) quantized version of [Qwen/Qwen3-8B](https://huggingface.co/Qwen/Qwen3-8B), created using [llm-compressor](https://github.com/vllm-project/llm-compressor) with calibrated quantization. ## Overview | Property | Value | |:---|:---| | **Base Model** | Qwen/Qwen3-8B | | **Parameters** | 8.19B | | **Quantization** | INT8 (W8A8) | | **Format** | `compressed-tensors` | | **Tool** | llm-compressor | | **Disk Size** | ~9.4 GB (2 shards) | ## Intended Use Quantized text encoder for [Flux 2 Klein 9B](https://huggingface.co/black-forest-labs/FLUX.2-klein-9B) image generation pipelines. Architecturally identical to the Klein 9B text encoder. ## Quantization Details - **Scheme**: W8A8 — 8-bit integer weights and activations - **Targets**: All `Linear` layers (excluding `lm_head`) - **Calibration**: 256 samples from C4, sequential pipeline with CPU offloading ## Hardware Requirements - **Minimum**: Any CUDA GPU with INT8 tensor core support - **Fallback**: Dequantizes to BF16 on unsupported hardware

Note9 oct.huggingface.co/vistralis/Qwen3-8B-INT8

Source context: 52 downloads · 0 likes · Pipeline text-generation · Library transformers · Repo vistralis/Qwen3-8B-INT8

Note31 juil.huggingface.co/vistralis/Qwen3-8B-INT8

Source context: 20 downloads · 0 likes · Pipeline text-generation · Library transformers · Repo vistralis/Qwen3-8B-INT8